Understand user intent
AI interprets the meal request and suggests relevant meal ideas and ingredients.
About This Demonstration
Duke Greens demonstrates how modern Java applications can use generative AI to solve a real business problem: translating what a user wants to eat into a shopping basket the application can validate, explain, and refine.
Instead of simply adding a chatbot, this demo shows how AI can create business value by helping enterprise applications bridge the gap between natural language intent and structured product data.
The business problem
Customers often know the outcome they want, but not the exact products they need. In grocery shopping, that can sound like:
“I want something healthy for four people that takes less than 30 minutes to prepare.”
Businesses, on the other hand, operate on structured data such as products, prices, basket contents, and order rules. Duke Greens shows how AI can interpret the request while the Java application keeps control of the business logic and trusted data.
How the demo works
Apply business rules
The Java application validates products against the curated catalogue and keeps authoritative data under application control.
Create a reviewable basket
The application assembles a virtual basket the visitor can inspect before anything is confirmed.
Keep humans in control
The visitor explicitly confirms the simulated order. AI assists the journey, but it does not complete consequential actions on its own.
Our AI approach
- Start with a measurable business outcome, not a model.
- Keep product, pricing, policy, and transaction data authoritative in the application or approved enterprise systems.
- Give AI bounded responsibilities and validate its outputs before business action.
- Design for evaluation, monitoring, and iterative improvement.
- Keep a human in control for consequential decisions and state changes.
What this demo is showing
Duke Greens is a technology demonstration, not a production shopping service. It illustrates how modern Java and generative AI can work together to create an intuitive, intelligent retail experience without processing real purchases or fulfilling grocery orders.
What the demo does not do
- It does not accept payments.
- It does not create a real order, reserve stock, arrange delivery, or create a customer account.
- Product availability, prices, and recommendations are illustrative and may not reflect real-world conditions.
Why this matters
The same pattern applies anywhere users describe an outcome instead of selecting a fixed product or service. AI can help interpret intent, suggest options, and explain choices while the application retains authority over the data and the workflow.
That makes the demo useful as a reference for retail, travel, insurance, internal assistants, and other enterprise journeys where the goal is not just automation, but better decision-making and a better user experience.
Your information
- Please do not enter personal, confidential, sensitive, health, payment, or account information.
- Keep requests limited to general meal-planning preferences.
Technology at a glance
Oracle Java 25 · Spring Boot 4.1 + Spring AI · Thymeleaf + HTMX · Deployed on Oracle Cloud Infrastructure (OCI)
Try the pattern or discuss your opportunity
Try the demonstration to see how Duke Greens translates a meal request into suggestions while the application keeps trusted data and confirmation in your control.
If you have a Java challenge or an AI idea worth testing, discuss the opportunity with Oracle Value Engineering.
Meal ideas are illustrative and are not medical, dietary, allergy, or nutritional advice.